Sitemap

A Day in the Life of an Architect in the Gradient World

20 min readAug 5, 2022
Gradient Worlds. Images generated with AI tools. Credits: Iro Armeni.

I envision a world where our spaces will support our evolving living, personal, and work needs in much more integrated, imaginative, and seamless ways than today because they will enable gradient realities. They will blend our physical and digital environments into one coherent and intuitive experience, with the blend taking place in different degrees of permeability between the two environments as needed by each situation. This article highlights some of the functionality of gradient realities and explores the rich design task of architects and their clients. Gradient realities will create much more reusable built spaces that are from the very beginning better suited to their users, as well as provide people with more equitable living conditions, hence achieving a more sustainable and inclusive built environment. Gradient realities will enable immersive connectivity, communication, and interaction with people on a global scale, as if they were co-located. Gradient realities will tap into our imagination and allow us to experience the world from many different perspectives, be that very individualized or very public. Join me in this article on a trip in the exciting future of a gradient world.

Paul glanced at the wall and noticed that the painting of Camille Pissarro “The Boulevard Montmartre at Night” that was there yesterday was changed for one of Claude Monet’s Water Lilies Series. The Museum of Art has dedicated this week to Impressionists and today was Monet Monday. As he steps closer to the painting while sipping his morning coffee, the painting comes to life; he sees the water surface glistening under the French sun and rippling under a calm summer breeze. He hears the leaves on the nearby trees rustling and some bees occasionally passing by. He smells the fragrant aromas of the lilies as the breeze blows in different directions and even feels that breeze on his clothing. Paul experiences all this thanks to advances in wearable computing that have made computers invisible; mixed reality enabling contact lenses, smart clothing, tiny speakers and earphones, and more. These advances allow him to exist in and experience all gradients of reality, from purely physical (real reality) to purely digital (virtual reality) and anything in between (e.g., augmented reality, substitutional reality, diminished reality, and augmented virtuality).

Even though he was taught this in Architecture School in the Art History class, he cannot remember the year the painting was created. “Show me more information about the painting” he says and a short description appears — it was created sometime between 1896 and 1926. There is more information to explore; visualizations and explanations of how Monet’s style evolved, the sociocultural conditions that pushed artists to Impressionism, and connections between Monet’s paintings to other artists, both artistically and geographically. He could even visit Monet’s garden at Givenry and walk around to explore it. Paul does not have the time for that today, he has to start working since he wants to see the results of the simulator on his new building design. However, the colors in the painting really resonated with him — you never know when inspiration will hit. He uses a digital paint palette and samples some of the main colors from the painting to use in one of his designs.

Sampling colors from Monet’s painting.
Paul looking at the Monet painting projected on the wall and sampling colors. Images generated with AI tools. Credits: Weixuan Lu (left), Eleni Alexandraki (right).

Before stepping into his office, he needs to tidy up the place. Last night his wife and twin children decided on the spur of the moment to play hide-and-seek and have a picnic in a medieval castle. This scene was not their decision; they used the “surprise me” mode of their apartment, which suggests activities and places that are out of the norm for each user’s behavioral patterns. He can see the digital toys of his children laying around, as well as the aftermath of the picnic — dishes, glasses, and napkins on the floor. They even rearranged the furniture to create more nooks and crannies that could serve as hiding places while playing in the castle. They substituted the layout of physical objects and building elements in their current modern apartment for those digital ones that resemble a medieval castle. The physical shape of the room changed as well to better follow the shape of the castle, with the ceiling coming down to form an arch. How is this possible, you might ask? Well, Paul lives in a “new-era” building, where the space adapts to different stimuli and needs, be that physical or digital. The shape and appearance of the room were not the only things that changed, the temperature in the room also dropped and the air felt more humid. Even their voices echoed from the medieval walls.

A few friends of the twins, located in completely different parts of the world, joined the game from the comfort of their own homes. Some joined in pure digital reality, while others used a gamut of gradient realities. Where the building elements and objects in the house coincided with the layout of the medieval castle, they were substituted by their digital medieval equivalent. The rest of the layout was either compensated by adapting the physical space of the room — when the user lived in a “new-era” building — or had only a digital presence. Paul met for the first time some of the twins’ friends. They joined from France, Mongolia, Namibia, and South Korea. Language was not a barrier these days, since translating systems have long been able to not only correctly translate to another language but also capture nuances originating from culture and difference in generations. He is seeing children growing up in the gradient world to be more open, unbiased, and empathetic to people from any ethnicity than generations of children that grew up in the old era. With his thoughts wandering around, Paul started tidying up. Cleaning up the digital space is quite easy — he will just hide the messy state from appearing in his view. This digital version of the space will remain there if they want to revisit it later, but they also have the option to visit any previous state. They can even visit a future state of the room, whatever that would be. What he finds extremely powerful is that a user can have many digital spaces, in varying degrees of digital completeness. These spaces can be accessed anytime from anywhere, be that in your own apartment or the other side of the world; the digital aspect of gradient spaces is always with you. Paul quickly focuses on tidying up the physical space; “living room mode” he says and the walls, floor, and ceiling return to their previous shape. He picks up the dishes, moves any furniture back in position, and he is ready to work.

Paul’s living room was converted to a medieval castle for a picnic with the family last night. Before (left) and after (right) the digital space gets hidden, revealing the physical space to tidy up. Images generated with AI tools. Credits: Eleni Alexandraki.

He steps into one of the extra rooms in the apartment that they treat as a multi-functional space — as an office, a gym, a space for arts and crafts, etc. He could stay in the living room, but today the twins will be back early from school and they might want to continue their hide-and-seek game with their friends. The physical space in the spare room is mainly empty, apart from a chair and a horizontal surface extruding from the wall at table height. He quickly enters office mode: screens, buttons, notepads, and tools, all virtual, appear in his sight. On the horizontal surface, Paul places a glass of water somewhere next to the digital keyboard. After all, he has no fear that the keyboard will get ruined if the water spills. The simulation result also appears. He has been assigned by a client who is a developer to renovate an “old-era” building into multifamily housing. The building was created with old design principles that did not take into account true user feedback or the possibility to exist in multiple realities. These days architects are able to design customized solutions for people based on their true interactions with the built environment. This allows them to create truly suitable, sustainable, inclusive, and adaptive spaces that can support any gradient reality, as well as the current and future, physical and digital needs of the users. Paul’s apartment was designed with these principles in mind, so he has experienced first hand the benefits of this new design paradigm.

The design process has been liberated from being a guess work on how to best design spaces and is now governed by quantitative and data-driven approaches. The aesthetics still remain highly on the architect, but even there Paul is now equipped with more tools to explore different alternatives. At the core of everything is a system that can simulate and predict how different users would live over time in a given space. Given the outcome of the simulation, the system can propose changes or alternate designs that would better fit the user. The user profiles, from which Paul can choose to populate the spaces in simulation, are also very diverse, ranging over many possible alternatives of culture, mobility, age, and more. Paul, as well as every designer, is finally getting feedback on his assumptions at the design time from actual user data and, therefore, has a much better chance of designing the right building. The system does not decide based only on the layout of the building, but takes into account structural, construction, energy, legislation, and circular economy considerations, as well as the desires of the clients. The latter is provided by tools created for easy and intuitive exploration, so that clients can figure out what they truly want in a space. The output of such tools has proven to be a better initialization of the design process, since it has been consistently providing faster convergence to the final design.

Paul looks at the simulation results (left, middle) and client avatar interacts with the design over time (right). Images generated with AI tools. Credits: Weixuan Lu (left, middle), Aleksandra Vorozheikina (right).

For this project, Paul and the rest of the engineering and construction team began by acquiring an as-is 3D semantic building model, which contains the current building elements and their properties (geometry, material, function, state, etc.). There is a global database that contains this information, up-to-date, for all existing buildings in the world. From this model, they were able to acquire the performance of all important variables of the building. Before deciding on any changes and engaging in more discussions with the client, Paul automatically computed the reusability scores of the building. Reusability scores are now at least equally as important as the energy efficiency scores. These are values that everyone has to calculate about their space, with the government imposing a lowest threshold of permitted reusability and providing financial incentives to owners for creating highly reusable spaces. Spaces with high scores are very attractive, since reusability reflects the ability of a space to adapt to future physical and digital needs of a diverse set of users. As Paul expected, the reusability scores of this building are close to the lowest threshold, which means that a more rigorous renovation process would be required.

Since Paul anticipated the low score, he had already asked the client to use the interactive tool to find a set of “wants” and “don’ts” for the new space. Although he could have used this input to make his own changes on the building, he opted for using the simulation system directly. After all, the system has learned over time how Paul prefers to approach design. Of course, there is always the possibility of the “surprise me” mode. Paul selected the user profiles based on the potential occupants of this building, which are a combination of couples with no children, families with babies or young children, and other families with adult children. In addition, the building is located in a neighborhood where historically there has been a strong mix of western and eastern cultures, particularly those of central Europe and Japan. Hence, Paul made certain to include user profiles with such ethnic diversity since it would have an impact on the final design. He also made sure to enable the search for supplying and acquiring material from the circular economy database. Lately, there has been a lack of wooden structural elements in the database that this building contains in abundance. The simulator was left to run overnight, although it takes less than 8 hours to finish the computations. It analyzed multiple solutions by simulating the interactions of the occupants with the building over two decades, such as their daily life in the initial state, future requirements with respect to working from home, new children being born in the families, and adults becoming elders. This means that both the building and its users will have evolved over time to have different needs and constraints (e.g., maintenance or other hardware and software updates for the building and mobility requirements for users as they age). The ability to construct the potential designs in an environmentally friendly manner was also part of the process, with the system providing different construction details depending on each design proposal. The system also has the capability to generate a safe and on-time detailed construction schedule by simulating the different stages of construction with high fidelity. This has drastically decreased, if not almost eliminated, injuries and accidents in construction sites.

As Paul is checking the suggestions of the simulation system, he sees that the structural engineer left comments that they would discuss in a few minutes during their meeting. What was important to the client is that the building would exist in the physical world with as minimal a physical space as possible, so that it can exist in many other realities simultaneously and take any shape and form the users would like it to have. Specifically, the client wanted to provide the users with two gradient realities. In addition to the pure physical space, the space needs to provide the experience of a mountain cabin with a view of the Alps and of a hacienda in Mexico. This means that Paul and the rest of the team should also design these other realities, so that they blend in an immersive and intuitive way with the underlying physical world. The users will have full control of how much of these two gradient realities (or any other they possess) will permeate the 100% physical space. It is also possible to have more than one digital world permeating into the physical at a given time. For example, the bedroom can be in the mountain cabin, whereas the living room in the hacienda. To create the digital worlds, they will start by using a style transfer tool, which takes characteristics from one architectural or building style and injects it to the target space. From this stage, they will make their own changes if the result is not satisfying.

Gradient Worlds, where space changes visually and/or functionally to adapt to user needs and desires. Images generated with AI tools. Credits: Austin Kim.

This comes in contrast to another project on which Paul is working. Although it is a “new-era” building, i.e., a new building designed and constructed for the gradient world, the client prefers to mainly live in an 100% physical space. Of course, there is some level of ubiquitous digitalization, but the interaction and blend of the physical and the digital is minimal. Specifically, the client wants the building to be almost identical to the one of his grandparents. Although the latter does not exist anymore and it was not part of the global database, Paul was able to restore most of the layout and design from pictures of the client spending his summers there during his childhood. He used a dedicated software that, using computer vision, is able to recreate a space with very high geometric and appearance accuracy from a set of images taken throughout a period of several years. It is in this project that Paul will use the color palette he “borrowed” from Monet’s painting this morning. The living room in the pictures was purple and Paul thought that the purples, blues, and greens in Monet’s painting would bring this space to life. Interestingly, even though this space resembles in many aspects an “old-era” lifestyle, it is still highly reusable. This is again possible thanks to the simulation system that is able to compute a design and construction plan that allows the building to exist in and support gradient realities and ever-shifting user needs.

The ability of physical spaces in the gradient world to become almost anything, from the perspective of an architectural style, view, function, etc., has allowed people to live in their dream space, regardless of whether it is, in the physical world, an architectural marvel or a box building. It has also created more equitable living conditions since there is a basic level of conditions available to everyone. From this base configuration, users can “build on” the space they want. This could mean making modifications on the base configuration before being built to meet their current needs or as time passes to meet evolving ones. It has also helped in decreasing the spread of urbanization over crop land, by shifting the paradigm from designing and building mono-semantic 100% physical spaces to multi-use gradient spaces, hence decreasing the physical housing footprint.

Thinking about the colors of that living room, Paul’s mind wandered to his own grandparents. They were proponents of the gradient world and early-on understood how beneficial it would be for them to make such a technology part of their everyday lives. Just last night Paul helped his grandmother, who lives 200 km away, to switch the battery on her hearing aid. Paul guided his grandmother’s hands as if he were performing this task, and in a matter of seconds the hearing aid was ready. Before, this would have taken them a few hours of back and forth phone calls where Paul would be trying to explain to her what to do, with very little success. Now, he can help them in many tasks and chores around the house even though he is located remotely. He and the twins can also spend a greater amount of quality time with them, since physical distance is not anymore a barrier in face-to-face immersive communication and interaction. Through this increased connectivity, they build on important family memories that will last a lifetime.

Paul helps his grandmother with her hearing aid. Image generated with AI tools. Credits: Aleksandra Vorozheikina.

Paul recollected his thoughts back to the renovation project. There is one particular apartment in this project that will be occupied by the client. He has made sure to choose the closest user profile to the client for the simulation and will showcase the design proposals to the client using that profile. The client will not only be immersed in the design proposals, but will also be able to see an avatar of themselves living in this space and interacting with the objects and building elements as if it were them. This has not only allowed clients to understand and participate in the design process more proactively, it has also decreased the amount of client-induced design changes during construction.

Paul will soon meet with the structural engineer. They will collaborate remotely on the simulation outcomes. They will look at the different visualizations and might spend some time taking a closer look at some of the simulation time intervals to get a clear understanding of the design’s suitability. They might even make some quick changes on the top three proposals to explore the design space in more depth. Until the meeting starts, Paul looks out the window and takes a glimpse at the construction site across the street. He always enjoys spending a bit of time observing the almost silent dance of the robots and humans collaborating to build in this new gradient world era.

Paul glances out of the window.
Paul glances out of the window. Images generated with AI tools. Credits: Weixuan Lu (left), Aleksandra Vorozheikina (right).

Reflections

There are still several steps to take in order to truly achieve gradient realities in the built environment in the way described in this article. The most fundamental step is to enable the seamless interaction of the physical and digital worlds on the entire spectrum of gradient realities. But this is not trivial.

First of all, the two worlds need to know their current state and communicate it with each other at any given point. Although this is simpler to solve from the digital world side, it becomes more complex when looking into the physical world. We can obtain this understanding in the physical realm with the use of Computer Vision and Machine Learning algorithms that predict the as-is status of a space using mainly visual data. Although progress in this field is substantial, we are yet to create systems that can robustly work in the real world. Given how achieving 100% accuracy with learning systems at any given condition is almost impossible, we should consider different ways that would allow the seamless interaction even if the state of the physical world is not entirely known or may contain incorrect information (e.g., multi-modal data sources, continuous learning, user in the loop, one/few shot learning, minimum viable information to exchange, etc.). Communication of the state and of relevant information between the two worlds is also a topic that requires further definition. What information is needed for each interaction? What is the structure of the information exchanged? How much of it is interpretable by humans? In addition, the ability of systems to not only know the current state of the two worlds, but also to predict next possible states given an interaction is highly important, since it will allow for safer and more suitable environments for users. Last but not least, the experience and interaction of users with the two worlds and their connecting systems is of utmost importance, in order for gradient realities to be part of user lives in an intuitive and seamless manner. Hence we should perform research and experiments to define best design principles, methods, and tools for creating gradient spaces over the entire spectrum of realities, when the user interface is not limited to a screen but can become the entire gradient space.

The above paragraph discusses the software advancements we need to make. However, there is much research to take place in the hardware side of gradient realities. To achieve a seamless integration in users’ lives, mixed reality devices and wearable computing are fundamental. We cannot achieve this future with clunky and tethered devices — we need to develop light-weight and invisible technology that will become part of our bodies. Our spaces should also be equipped with sensing devices, from visual ones to others that measure temperature and humidity. In addition, to minimize the perceptual gap between the digital and the physical, we require powerful continuous rendering hardware. To further minimize this gap, a realistic rendering is not enough. We should consider how the two worlds react to each other, and specifically how the physical world reacts to digital stimuli. It can be as minimal as lowering the lights and closing the blinds or can result in larger transformations per principles of kinetic designs (e.g., partial movement of a floor and walls to create a meeting space with sitting and working surfaces). Especially for the latter to happen, we should consider which construction methods and materials enable this, while remaining in line with a sustainable and reusable future. Regardless of whether we are discussing hardware or software, we need to clearly define interoperability across different devices, brands, operating systems, and new versions of them. This is something with which we are currently struggling and — despite market-based incentives — the future of gradient realities will not exist if our systems cannot continuously and effortlessly communicate with each other.

It would be remiss of me not to discuss the paramount importance of addressing privacy and ethics concerns when dealing with data that contain highly sensitive information (e.g., humans and their personal effects). It should be at the forefront of our considerations; from the use of data collection and processing systems to the development of algorithms that provide appropriate handling of this information. For example, we should consider developing and employing sensing devices that can locally process information and create lower risk representations of observations without storing raw data. This would require lightweight and fast methods that allow on-the-fly processing. Without establishing clear and transparent boundaries on the type and way of collecting, storing, using, accessing, and sharing data, we cannot expect that people on the other end of our systems — and therefore gradient spaces — will trust, support, and collaborate.

The future of a gradient world gives us the unique opportunity to reconsider the way we are designing our physical and digital worlds. To address sustainability concerns, among other efforts, we should be designing and (re-)building spaces that are able to support the evolving living of users over the years. But how can we create spaces that provide such support, or else that are suitable for users over time? In today’s architectural practice we design for users but without having true feedback of how users interact with spaces and what changes they make in order to make a space more suitable to them. We need to understand what makes a space suitable for diverse user profiles, and provide methods and tools to architects to incorporate this information into their designs. In the article above, I mentioned a simulation environment that allows architects to explore the vast design space of gradient realities and create design proposals that are suitable for users. This environment can serve as a co-pilot that will allow designers to fully navigate the design space while placing the human in the center.

Building such a simulation environment is not straightforward and there are many problems that need to be understood and solved before achieving it. Two main characteristics this environment should have are the ability to represent (a) the different physical and digital states of our environments with high fidelity, as well as (b) human interactions with the environment in a realistic manner. The latter does not mean visually rendered so that humans can perceive them as realistic; it concerns the type of interactions with the environment, their cause and effect, as well as the diversity of the user models they represent. To build such a simulation environment we need to observe diverse users interacting with diverse environments over time, and develop algorithms that analyze this data, model the environments and interactions, simulate them, and have the ability to predict possible interactions and/or environments given a user model. When the encoded interactions represent diverse user models (e.g., culture, mobility, and age), simulating a user’s interactions with a new design can inform us of the latter’s suitability to the user, thus resulting in a more inclusive built environment or in one for which we understand its limitations and the effort required to adapt it to a different user model.

In addition to this simulation environment, designers will also have access to gradient reality tools to design and communicate in an intuitive manner. We should identify how these tools will look and particularly how they can enhance the capabilities of their users.

Gradient realities can be an exciting future that has the potential to shape our built environment in an unprecedented way. However, we are currently focusing on a very narrow scope of this future, such as on how we can use virtual and augmented reality tools for design applications or on creating new buildings and their realistic renderings for the Metaverse. We should reframe our view of the gradient world to the one suggested in this article, and — as architects, engineers, and builders of the physical and digital worlds — join the tech companies in (re)shaping this future.

Acknowledgments: This article reflects my own views and thoughts on a possible future scenario for the built environment, that of gradient realities. However, I would not have been able to imagine it and put it into words if it weren’t for the inspiring work I witnessed and the discussions I had over the past nine months with colleagues at ETH Zurich, Stanford University, and beyond. I will refrain from naming people to avoid omissions, but I will highlight one in particular that has been my sounding board and mentor over the past 8 years. I would like to explicitly acknowledge Martin Fischer from Stanford University for his continuing support and for lending me his imagination to explore ideas, regardless of how out-of-the-box or not they might be.

If you wish to cite the article:

@misc{Armeni_2022, title={A day in the life of an architect in the gradient world}, url={https://medium.com/@iroarmeni/a-day-in-the-life-of-an-architect-in-the-gradient-world-70c1996710fe}, journal={Medium}, publisher={Medium}, author={Armeni, Iro}, year={2022}, month={Aug}}

--

--

Iro Armeni
Iro Armeni

Written by Iro Armeni

Prof@Stanford CEE | Prior: PostDoc@ETHZ, PhD@ Stanford. Researching Perception, AI, & Computer Vision to transform Design, Construction & Facility Management

Responses (1)